Car crashes have killed 3x as many people in the US as all wars:

Sources:

• Motor vehicle deaths in U.S. by year
• United States military casualties of war

The code below generates this plot:

from datetime import datetime as dt
from os.path import basename, exists
import pandas as pd
import plotly.express as px
import re
from urllib.request import urlretrieve

Helper for:

• scraping a URL (in this case, a Wikipedia article)
• caching it locally
• parsing any tables it contains
• returning the table at index idx

def get_table(url, idx, path=None):
    if path is None:
        name = basename(url)
        path = f'{name}.html'
    if not exists(path):
        print(f'Downloading {url} to {path}')
        urlretrieve(url, path)
    tables = pd.read_html(path)
    table = tables[idx]
    return table

1. U.S. Motor Vehicle Fatalities

Pull table from Wikipedia article: Motor vehicle deaths in U.S. by year

cars_url = 'https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in_U.S._by_year'
cars_path = 'Motor_vehicle_fatality_rate_in_U.S._by_year.html'
cars = get_table(cars_url, idx=0)
cars

	Year	Deaths	VMT– Vehicle miles traveled (billions)	Fatalities per 100 million VMT	Population	Fatalities per 100,000 population	Change in per capita fatalities from previous year
0	1899	26[5]	NaN	NaN	NaN	NaN	NaN
1	1900	36	NaN	NaN	76094000.0	0.05	NaN
2	1901	54	NaN	NaN	77584000.0	0.07	47.1%
3	1902	79	NaN	NaN	79163000.0	0.10	43.4%
4	1903	117	NaN	NaN	80632000.0	0.15	45.4%
...	...	...	...	...	...	...	...
116	2015	35485	3095.0	1.15	321370000.0	11.06	7.6%
117	2016[8]	37806	3174.0	1.19	323121000.0	11.59	4.8%
118	2017[8][9]	37473	3213.0	1.16	326213213.0	11.40	-1.6%
119	2018[9][10]	36560	3223.0	1.13	327096265.0	11.18	-1.9%
120	2019[11][12]	36120	NaN	1.10	NaN	NaN	-1.2%

121 rows × 7 columns

Clean data:

• remove citation brackets
• convert data types, index by year
• add 2020 stat (38,360; source)

for k in cars:
    if cars[k].dtype == 'object':
        cars[k] = cars[k].str.replace('(?: ?\[\d+\])+$', '')
cars = (
    cars
    .astype({'Year': int, 'Deaths': int})
    .set_index('Year')
    .Deaths
)
cars.loc[2020] = 38360
cars

Year
1899       26
1900       36
1901       54
1902       79
1903      117
        ...  
2016    37806
2017    37473
2018    36560
2019    36120
2020    38360
Name: Deaths, Length: 122, dtype: int64

2. U.S. casualties of war

Pull table from Wikipedia article: United States military casualties of war

wars_url = 'https://en.wikipedia.org/wiki/United_States_military_casualties_of_war'
wars_path = 'United_States_military_casualties_of_war.html'
wars = get_table(wars_url, idx=0)
wars

	War or conflict	Date	Total U.S. deaths			Wounded	Total U.S. casualties	Missing	Sources and notes	Deaths as percentage of total population
	War or conflict	Date	Combat	Other	Total	Wounded	Total U.S. casualties	Missing	Sources and notes	Deaths as percentage of total population
0	American Revolutionary War	1775–1783	8000	17000	25000	25000	50000	NaN	NaN	NaN
1	Northwest Indian War	1785–1796	1,056+	NaN	1,056+	825+	1,881+	NaN	[1][2][3]	NaN
2	Quasi-War	1798–1800	20	494[4]	514	42	556	NaN	[4][5]	NaN
3	First Barbary War	1801–1805	35	39	74	64	138	NaN	[6][7]: 25–27 [8][9]	NaN
4	Other actions against pirates	1800–1900	36	158+[7]: 42	194+	100+	294+	NaN	[5][10][11][b]	NaN
...	...	...	...	...	...	...	...	...	...	...
77	War in Afghanistan	2001-2021	1,833 [83]	385[84][83]	2218	20,093[83]	22311	NaN	[f][83][84]	NaN
78	Iraq War	2003–2011	3,836*	961	4497	32222	36710	3	[83]	NaN
79	Intervention against the Islamic State of Iraq...	2014–present	16	60	76	81	157	NaN	[83]	NaN
80	Raid on Yemen	2017	1	NaN	1	3	4	NaN	[85]	NaN
81	Total	1775–2019	666,441+	673,929+	1,354,664+	1,498,240+	2,852,901+	40,031+	NaN	NaN

82 rows × 10 columns

Clean data

• flatten column index
• scrub citation brackets and trailing ~/+ characters
• convert to numbers, drop na's, fill in "present" to be the current year

wars.columns = [ col[1] for col in wars.columns ]  # Flatten column index
for k in wars:
    if wars[k].dtype == 'object':
        wars[k] = wars[k].str.replace('(?: ?\[\d+\](?:\:\s?\d+(?:-\d+)?)?)+$', '')
        wars[k] = wars[k].str.replace('[~+]$', '')


wars = wars.rename(columns={'War or conflict': 'War'}).set_index('War')[['Date','Total']]
wars = wars.dropna(subset=['Date', 'Total'], how='any')
wars.Total = wars.Total.str.replace(',', '')
year = dt.now().year
wars.Date = wars.Date.str.replace('present', str(year))
wars

	Date	Total
War
American Revolutionary War	1775–1783	25000
Northwest Indian War	1785–1796	1056
Quasi-War	1798–1800	514
First Barbary War	1801–1805	74
Other actions against pirates	1800–1900	194
...	...	...
War in Afghanistan	2001-2021	2218
Iraq War	2003–2011	4497
Intervention against the Islamic State of Iraq and Syria	2014–2021	76
Raid on Yemen	2017	1
Total	1775–2019	1354664

77 rows × 2 columns

Parse dates into start and end years, and numeric casualty ranges into lo and hi estimates

def parse_ints(years, min_key, max_key):
    matches = re.findall('\d+', years)
    years = [ int(year) for year in matches ]
    m = min(years)
    M = max(years)
    return {min_key: m, max_key: M}

wars = pd.concat(
    [
        wars, 
        wars['Date'].apply(parse_ints, min_key='start', max_key='end').apply(pd.Series),
        wars['Total'].apply(parse_ints, min_key='lo', max_key='hi').apply(pd.Series),
    ],
    axis=1,
)
wars

	Date	Total	start	end	lo	hi
War
American Revolutionary War	1775–1783	25000	1775	1783	25000	25000
Northwest Indian War	1785–1796	1056	1785	1796	1056	1056
Quasi-War	1798–1800	514	1798	1800	514	514
First Barbary War	1801–1805	74	1801	1805	74	74
Other actions against pirates	1800–1900	194	1800	1900	194	194
...	...	...	...	...	...	...
War in Afghanistan	2001-2021	2218	2001	2021	2218	2218
Iraq War	2003–2011	4497	2003	2011	4497	4497
Intervention against the Islamic State of Iraq and Syria	2014–2021	76	2014	2021	76	76
Raid on Yemen	2017	1	2017	2017	1	1
Total	1775–2019	1354664	1775	2019	1354664	1354664

77 rows × 6 columns

Use the low-end estimates, drop final "total" row

wars = (
    wars
    .iloc[:-1]
    .copy()
    [['start', 'end', 'lo']]
    .rename(columns={'lo': 'deaths'})
    .sort_values('end')
)
wars

	start	end	deaths
War
American Revolutionary War	1775	1783	25000
Northwest Indian War	1785	1796	1056
Quasi-War	1798	1800	514
First Barbary War	1801	1805	74
Chesapeake–Leopard affair	1807	1807	3
...	...	...	...
Iraq War	2003	2011	4497
Raid on Yemen	2017	2017	1
Colombia	1994	2021	8
War in Afghanistan	2001	2021	2218
Intervention against the Islamic State of Iraq and Syria	2014	2021	76

76 rows × 3 columns

Approximate each war's deaths as being evenly spread over the years in which the war was waged:

def spread_war(r):
    start, end, deaths = r['start'], r['end'], r['deaths']
    years = range(start, end + 1)
    num_years = len(years)
    deaths_per_year = deaths / num_years
    return [ dict(war_deaths=deaths_per_year, year=year) for year in years ]

war_deaths_per_year = (
    wars
    .apply(spread_war, axis=1)
    .explode()
    .apply(pd.Series)
    .astype({'year': int})
    .reset_index()
    .groupby('year')
    .sum()
)
war_deaths_per_year

	war_deaths
year
1775	2777.777778
1776	2777.777778
1777	2777.777778
1778	2777.777778
1779	2777.777778
...	...
2017	116.404762
2018	115.404762
2019	115.404762
2020	115.404762
2021	115.404762

239 rows × 1 columns

3. Combine per-year car and war deaths

deaths_per_year = pd.concat(
    [
        war_deaths_per_year['war_deaths'].rename('War'),
        cars.rename('Cars'),
    ],
    axis=1,
)
deaths_per_year

	War	Cars
1775	2777.777778	NaN
1776	2777.777778	NaN
1777	2777.777778	NaN
1778	2777.777778	NaN
1779	2777.777778	NaN
...	...	...
2017	116.404762	37473.0
2018	115.404762	36560.0
2019	115.404762	36120.0
2020	115.404762	38360.0
2021	115.404762	NaN

245 rows × 2 columns

Fill in missing years, add cumulative sums

start, end = deaths_per_year.index.min(), deaths_per_year.index.max()
years = pd.Index(range(start, min(2020, end) + 1), name='year')
deaths = years.to_frame().merge(deaths_per_year, left_index=True, right_index=True)[['War', 'Cars']].fillna(0)
deaths['War (Total)'] = deaths['War'].cumsum()
deaths['Cars (Total)'] = deaths['Cars'].cumsum()
deaths

	War	Cars	War (Total)	Cars (Total)
1775	2777.777778	0.0	2.777778e+03	0.0
1776	2777.777778	0.0	5.555556e+03	0.0
1777	2777.777778	0.0	8.333333e+03	0.0
1778	2777.777778	0.0	1.111111e+04	0.0
1779	2777.777778	0.0	1.388889e+04	0.0
...	...	...	...	...
2016	115.404762	37806.0	1.345876e+06	3720188.0
2017	116.404762	37473.0	1.345992e+06	3757661.0
2018	115.404762	36560.0	1.346108e+06	3794221.0
2019	115.404762	36120.0	1.346223e+06	3830341.0
2020	115.404762	38360.0	1.346339e+06	3868701.0

244 rows × 4 columns

4. Plot

fig = px.line(
    deaths[['War (Total)', 'Cars (Total)']],
    labels={
        'variable': '',
        'value': 'Deaths',
        'index': 'Year',
    }
)
fig.update_layout(
    title='U.S. deaths over time: war (combat + non-combat) vs. cars (crashes only)', title_x=0.5
)
fig

fig.write_image('car_war_deaths.png')